My study area contains a basin, but the model identifies the basin as a water area. I used TFA=1000,5000,10000, but the results are the same.
This is the resulting stream.tif:
This is the corresponding elevation data：
By a “basin” do you mean that area does not drain anywhere? I.e. a sunken area entirely surrounded by higher elevations?
SDR generally does not work very well on flat areas, since it works by modeling water flowing downslope. Please search the forum for previous discussions about this, for example Seasonal water yield, issue with quickflow/stream map
Thank you！Basin is a common geographical landform, which refers to a terrain area with high terrain around and low terrain in the middle. The basin in my research area is almost all desert and saline-alkali land, and it rarely rains. The soil retention here is very poor. However, the model identifies this basin as a water area (in fact, it is just a depression full of desert).
Thanks for the clarification here @lingyun, and I’m sorry for the delay here. The model uses the threshold flow accumulation value to determine what it considers a stream, so since this large, flat area is one of the lowest-elevations parts of the DEM, the model determines it to be a stream. This is compounded by the multiple-flow-direction routing algorithm, which can result in large, flat areas like this all being determined to be a stream.
As far as workarounds go, this comment from @swolny over in Riparian buffer values issues because of NDR output streams wider than in reality - #2 by swolny might be helpful. The post is about NDR, but SDR uses the same routing algorithm, so the approach should still work well.